bootc-dev · nimbinatus · Aug 7, 2026
diff --git a/meetings/2025-07-11.md b/meetings/2025-07-11.md
@@ -0,0 +1,37 @@
+## 11 July 2025
+### Attendees:
+- Joseph Marrero Corchado (Red Hat, Inc.)
+- Colin Walters
+- Robert Sturla (Tesco Bank/Universal Blue)
+- Laura Santamaria (she/her)
+- Hristo Marinov
+- John Eckersberg (Red Hat, Inc.)
+- Dusty (he/him)
+- Antheas Kapenekakis (Bazzite)
+- Mohan
+- Chris Kyrouac
+- Gursewak Mangat
+
+### Agenda:
+ - New release status https://github.com/bootc-dev/bootc/issues/1390 
+  - folks agreed on this  
+ - [Laura] [KubeCon NA 2025](https://events.linuxfoundation.org/kubecon-cloudnativecon-north-america/) [Project Pavilion application](https://events.linuxfoundation.org/kubecon-cloudnativecon-north-america/features-add-ons/project-opportunities/#description-of-opportunities)
+   - November 10-13 - Atlanta, Georgia
+   - Who is going to KubeCon NA 2025 already?
+     - Laura
+   - We'll get a project pavilion submission scheduled (probably from Joseph or Colin who already submitted a talk)
+ - Ublue - Collaboration
+     - Folks are joining :)
+     - Robert, Antheas from the Ublue community
+     - Colin would like to do 1:1s
+         - dustymabe: an office hours like set time could help facilitate this (+1 - Laura)
+ - rechunker alignment https://github.com/hhd-dev/rechunk 
+ - progress-fd: https://github.com/bootc-dev/bootc/issues/1016 
+ - systemd-sysext frontend: https://github.com/bootc-dev/bootc/issues/7
+    - Motivated by combinatorial explosion of gnome|kde * nvidia|amd * surface|framework|lenovo
+    - discussion of downsides of systemd-sysext as defined today, vs  
+
+### TODO:
+* [x] Put project pavilion application
+* [x] Keep smaller, more focused meetings
+
diff --git a/meetings/2025-07-18.md b/meetings/2025-07-18.md
@@ -0,0 +1,47 @@
+## 18 July 2025
+### Attendees:
+- Laura Santamaria (she/her; Red Hat)
+- Hristo Marinov
+- Fernando Lozano
+- Colin Walters (he/him; Red Hat)
+- Joseph Marrero Corchado (Red Hat, Inc.)
+- Matteo Piccinini (n/a)
+- Robert Sturla (Tesco Bank)
+- Dusty Mabe
+- Chris Kyrouac
+- Gursewak Mangat
+- John Eckersberg (Red Hat, Inc.)
+- 
+
+### Agenda:
+- Release 1.5.1: {%preview https://github.com/bootc-dev/bootc/pull/1422 %}
+  - Thanks @robert!
+- [Laura] project pavilion update?
+  - Request form due Monday, shouldn't block on travel
+  - Let's do it live
+  - ✅
+- [Laura] Static site generation for landing page
+  - Review!
+  - {%preview https://github.com/bootc-dev/bootc-dev.github.io/pull/3 %}
+  - adding logos as examples: https://fedoraproject.org/coreos/; https://projectbluefin.io/
+- [Colin] QMU banned contributions from AI. Want to talk about it. Contribution policy?
+  - Thoughts?
+  - Require use of `<Assisted-by>` tag to identify model/tool
+  - Errant AI comment contributed to recent bug
+  - [John] +1 on attribution
+  - [Dusty] do we want to limit, or allow?
+    - [Colin] has a pretty big impact, so want to know what folks think
+  - [Joseph] note in readme or contributing section makes sense. Maybe have a bot that highlights that on PR? People probably won't say anything because part of workflow
+  - [Dusty] Can't prevent, but policy. Any examples of wasted time on clearly generated by AI PR
+    - [Colin] Can tell 90% of the time. Most modern foundational models love bulleted lists, so it's obvious. Kinda wacky to put md doc in top level of repo for PR
+    - [Laura] gave overview from OSPO and the other container group discussion
+  - [Colin] Will open a discussion
+
+### TODO:
+- [ ] Laura to explore adding logos to PR.
+- [ ] Laura to add GitHub Actions for publication
+- [ ] Look for info on domain handling for static site
+- [ ] Colin to open a discussion about the AI assisted PRs
+- [x] Laura to find and share the public Containers Cabal recording
+
+
diff --git a/meetings/2025-07-25.md b/meetings/2025-07-25.md
@@ -0,0 +1,46 @@
+## 25 July 2025
+### Attendees:
+- Robert Sturla (Tesco Bank / Universal Blue)
+- Hristo Marinov
+- Ben Breard
+- Laura Santamaria (she/her)
+- Joseph Marrero Corchado (Red Hat, Inc.)
+- Colin Walters
+- Dusty Mabe
+- Mohan Shash
+- Gursewak Mangat
+
+### Agenda
+- {%preview https://github.com/bootc-dev/bootc/milestone/5 %}
+    - Anyone have anything else to add to the milestone?
+    - Added a couple
+- Info about contributing: {%preview https://developers.redhat.com/blog/2025/07/23/shape-future-linux-contribute-bootc-open-source-project %}
+- [Colin] Assigning/delegating issues
+    - Right now not auto-assigning reviews and issues. Worth starting to do?
+    - Bot to round-robin reviews?
+    - https://docs.github.com/en/organizations/organizing-members-into-teams/managing-code-review-settings-for-your-team
+    - Thoughts, opinions, screaming fights?
+        - [Joseph] OpenShift repos have something automatic like this. Agree the custom stuff is heavy-handed for us.
+        - [Colin] Merge queue is a huge example of something that GitHub added that used to be custom for other projects
+        - [Colin] will take action item to set something up for this
+- [Laura] Website PR - build is technically running on my fork
+    - {%preview https://github.com/bootc-dev/bootc-dev.github.io/pull/3 %}
+    - https://nimbinatus.com/bootc-dev.github.io/ (except my domain stuff is messing with links)
+    - [Ben] Do we want to move the this week in bootc there?
+    - [Ben] What about https://containers.github.io/bootable/ ? Move it? Use it? Wipe it?
+        - [Colin] There's two things: Specs/standards vs how it works. People have asked for "what's the bootc spec?" would be good to add it somewhere... Probably don't squash them yet.
+        - Containers org has been catchall. Just transfer the repo into bootc-dev? Could explain the spec there.
+        - Worth linking or leave it separate?
+            - Maybe come back to it? Any strong opinions?
+            - [Mohan] Linking would be helpful. Website at bootc.dev that links to it will make it easier to find...
+    - [Colin] Let's land website and iterate from there
+- rollup of events/blogs/releases, picking back up BCTW?
+    - Duplicates news?
+    - If there are blog topics we need to write on, talk with Ben?
+
+### TODO
+- [ ] Colin - set up review round-robin bot
+- [ ] Laura - go fix the css links
+- [ ] Laura - cname setup
+- [ ] Laura - fix the double-workflow issue on the site
+
diff --git a/meetings/2025-08-01.md b/meetings/2025-08-01.md
@@ -0,0 +1,26 @@
+## 1 Aug 2025
+
+### Attendees:
+- Colin Walters
+- Mohan Shash
+- Jonathan Lebon
+- Robert Sturla
+
+### Agenda
+
+- bootloaders and composefs?
+  - Robert may look at WIP for detecting images without bootupd and defaulting to systemd-boot
+  - https://github.com/bootc-dev/bootc/issues/806#issuecomment-3145079372
+- composefs
+  - Live working session later 1:30pm EST
+  - future composefs will not require reprovisioning 
+  - Discussion of https://github.com/bootc-dev/bootc/pull/1471
+- coreos transitioning to bootc
+  - https://github.com/bootc-dev/bootc/issues/1441 
+  - https://github.com/coreos/rpm-ostree/issues/4994
+  - https://github.com/bootc-dev/bootc/issues/1320
+  - https://github.com/bootc-dev/bootc/issues?q=state%3Aopen%20label%3A%22area%2Fcoreos-alignment%22
+  - dnf discussion
+
+
+
diff --git a/meetings/2025-08-08.md b/meetings/2025-08-08.md
@@ -0,0 +1,43 @@
+## 8 Aug 2025
+
+### Attendees
+- Robert Sturla (Tesco Bank / Universal Blue)
+- Hristo Marinov
+- Laura Santamaria (she/her; Red Hat)
+- Colin Walters
+- Joseph Marrero Corchado (Red Hat)
+- John Eckersberg (Red Hat, Inc.)
+- Chris Kyrouac
+- Dusty Mabe (he/him; Red Hat)
+- Gursewak Mangat
+- Jonathan Lebon
+- Sean Thrailkill
+- afterstory
+
+### Agenda
+- [Colin] More on composefs {%preview https://github.com/bootc-dev/bootc/issues/1498 %}
+    -  [Colin] Clientside reworking on how we store files. Then there's the build side to this whole thing. What's the user experience look like for creating an image
+    -  [Colin] Recap from previous meeting
+    -  [Colin] Fatal flaw of binary copying into host
+    -  [Colin] Issue is about build side
+    -  [Colin] Logistically simplest - containers build however you want, bootc cleans up, then user provide secure boot key
+    -  [Jonathan] Not following super closely, but feels like we're starting with the hard case. Any work around non-sealed composefs builds today? Does it work?
+        -  [Colin] Yes, kinda. "It Depends":tm: Gets into whole problem around how we transition installs. Doesn't need to be sealed, but only target UKIs to start. We get into mechanical issue of default Fedora base images aren't set up for UKI. ostree backend does not really support UKIs (at least not with UEFI). Yes, it definitely will support nonsealed systems. Assuming we want to run this tool on your container images anyway.
+        -  [Jonathan] Need to make sure that, when designing, want to consider how migrate existing systems, and how will work with unsealed systems. Sealed case may be a specialized thing in beginning. May affect design; overconstrain.
+        - [Colin] We will definitely support transitioning existing systems. In end, can always boot without secure boot, too.
+        - [Colin] not committing to anything; this is still experimental. Nothing stops us from allowing flow within Dockerfiles, too.
+    - [John] Haven't really explored it a whole lot. May change, and likely will hit many walls. Just wanted to start discussion. Please feel free to comment in issue! Help find flaws in plan :)
+    - [Colin] Want people to feel free to ask questions in this meeting! Don't be afraid to ask for clarity, ask to help contribute.
+- [Sean] Recently gave talk at Flock about state of bootc! Big fan, want to get involved. We've shifted away from how installation process used to work. Used to be build your artifacts, now transitioning to replace the existing install, especially in more cloud-centric areas. What was the impetus for the change in direction?
+    - [Colin] Very interested. Wouldn't say it's backing away. We know we have to support a flow where you have a container image, then allow to turn into ISO or raw disk file. Too many use cases for that to give it up (e.g., edge, IOT devices, want to preconfigure before ship). We have to support that. Tension is in some clouds, managing disk images kinda stinks. Esp. AWS, as an example. There's two different worlds, and we do need to do both.
+    - [Dusty] bootc image builder supports uploading AMI for AWS, but usually you have to figure that part out on your own. Now also options to boot and then replace. That's more of a "let's remove the extra step from the user's responsibility, allow for existing thing that can be paved over." Might make it easier.
+    - [Sean] Makes sense. Sounds like this is to make it simpler for the user; pave over idea makes sense for the user.
+    - [Dusty] idea is they already have podman, which can run the container than can then pave your system. Bootc is on engineer's device. Positives and negatives. Positive: You start with whatever image for OS is on your cloud, and then you have to rebase that instance to your container. Negative: Feels like there's a new cloud that pops up everyday. Feels like we're constantly chasing the ball. Better experience is wherever you are is a starting point. So this allows to start from *something* you can start from. Doesn't even have to be EL-based. Can be Ubuntu. Can get to success without having to create your own disk image.
+    - [Colin] Not an installer in the partitioning sense. We want to make it easy/happy path for someone making own OS or distro - closer you can get to OS is container, then the installer area can be smaller. Bootc doesn't know how big you want your partition to be, for example, but if you can use whatever installer to install your containerized OS, then you're in a better space. Freya Labs also investigating bootc, called readymade.
+    - [Laura] I'll see if I can find them. (chat noted that they're in the universal blue discord)
+- [Jonathan] Choice of filesystem in container image and possibility of putting the choice of the filesystem in the image itself
+    - [Colin] Would like to support systemd-repart as part of MVP first, possibly able to support something like this later.
+- Lots of talk about new friends, inviting new friends, and timezone issues (we know this isn't great for Europeans and others.)
+- Discussion of recording meeting, general agreement we should probably record and post
+ - One option: Split 50%/50% recorded vs not, the advantage of this would be that the second half can be for "stupid questions" that people may not be comfortable having on the Internet Forever
+
diff --git a/meetings/2025-08-15.md b/meetings/2025-08-15.md
@@ -0,0 +1,62 @@
+## 15 Aug 2025
+
+### Attendees
+- Colin Walters
+- John Eckersberg (Red Hat, Inc.)
+- Joseph Marrero Corchado (Red Hat, Inc.)
+- Mohan Shash
+- Michael Fox
+- Preethi Thomas
+- Robert Sturla (Tesco Bank)
+- Hristo Marinov
+- Laura Santamaria (she/her; Red Hat)
+- Johnathan Lebon
+- Sean Thrailkill
+- Gursewak Mangat
+- Dusty
+- Michael Tunnell (Framework)
+- AfterStory
+- Jorge Castro (CNCF/Linux Foundation)
+
+### Agenda
+- [Jonathan] bootc plugin system
+    - [Jonathan] What do we want the UX to be like? Two high-level approaches: dnf drives bootc, or bootc drives dnf. This is currently Fedora focused, but implementation details is not Fedora focused. If dnf does persistent package layering, goal should still be bootc upgrade and bootc status work and that's the user. For that to work, bootc and dnf need to work together, hence plugin system. How do we have bootc communicate with other applications (this is where it's non-Fedora centric). Logic shouldn't likely live in bootc, right? What's the plugin API look like?
+    - [Colin] Wasn't there an issue about this?
+    - [Jonathan] context the first time was originally rpmtree
+        - {%preview https://github.com/bootc-dev/bootc/issues/337 %}
+    - [Colin] Complicated. Think about progress reporting, especially as GUI. We want a progress bar, coherent representation of what's happening. Want accurate change detection. Want things visually simple to understand for user. If bootc forks off other binary, punt the issue of how representation of state works. Inclination is to punt to the tool; make it the job of the higher-level tool.
+    - [Jonathan] Going to server model, you would do dnf status, for example, and dnf upgrade?
+    - [Colin] Yes, can't avoid the tools and need to know about both in some cases. Could cover a lot of cases if we taught the package managers how to manage a sysext that they own locally. There's already a lot of tooling to introspect sysext.
+        - {%preview https://github.com/rpm-software-management/dnf5/issues/1731 %}
+    - [Dusty] if we make it generic (e.g., local layering), local packages is one plugin that's really important. Being able to add a package that's not in the thing that you're following is extremely convenient. In most cases, there's a package manager tool that you could make plugins, and tool itself could learn how to do thing in sysext, then bootc doesn't have to care (or maybe it does?). How many things do people have to touch to upgrade system? What's the better user experience? If you step away from package managers/systems, then that's where it gets iffy. Might make more sense to make bootc the one that controls that instead of having dracut try to call bootc. Might be better bootc managing as plugin way.
+        - Bootc notices update, grabs it, then pipeline - output is input to next step. Output of pipeline is from plugins. Idea is more that mutiple plugins can stack. Or can combine to multistage build (yay pipelines).
+    - [Colin] Wary of adding more complexity to bootc right now. Issues in place, example: We don't have unified storage (podman build/docker build is super awkward at the moment). There's prepwork we need to add. Someone else might need to own that task. Still problem of non-package content.
+    - [Sean] Inclined to agree. Keep bootc simple; teaching other tech to know how to layer versus trying to edit the image that is actually booted (and security implications for that).
+    - [Dusty] Use case thinking about is in a scenario where someone is autoupdating their system. e.g., normally they don't touch it. How can they keep the things that they have done to keep everything up to date? That's why like the plugin model. Then other things get updated alongside the upgrade that happens (e.g., the user who ran a command years ago and doesn't remember that command). How to keep config that was made up to date, etc. Have same problem today in ostree world where there's package layering, e.g., orphaned packages.
+    - [Jonathan] Package layering is a big one as it's a common want. When you boot up a bootc system, you use bootc to manage it, from point on to add package, then you stop touching bootc and have to know other tools. Introduces split brain. Is bootc an implementation detail of dnf? Feels awkward. Would be better that bootc knows the state of the system. Make bootc the entrypoint to managing your system.
+    - [Dusty] Would be nice if someone runs bootc status and copy-paste to issue filed, then shows full system. Would be helpful to have single pane of glass.
+    - [Laura/Colin] Let's move to an issue for more discussion. Jonathan to make the issue.
+- [Mohan] Follow up - when is the site going live: https://github.com/bootc-dev/bootc-dev.github.io? 
+    - [Laura] The site is live at https://bootc-dev.github.io, but are you asking about moving it to bootc.dev?
+        - [Mohan] Thanks, I wasn't aware that it was live already :-)
+- [Robert] what info should we get from users who hit this? {%preview https://github.com/ostreedev/ostree/issues/2283 %}
+    - [Robert] One of the most annoying bugs we face. No easy way to resolve when it happens, no reproducable steps. What can we do to help debug?
+    - [Colin] Don't know offhand. Could ask folks to share journals and get more logs. AI version churning on that.
+    - [Hristo, in chat] {%preview https://discussion.fedoraproject.org/t/silverblue-wont-boot-after-forced-shutdown-searching-for-ways-to-recover-it/155432 %}
+    - [Jonathan] Catchall issue. Symptoms look the same, but not same root cause. When initially filed this, cause was people doing reinstalls and they had old install on disk one but new on disk 2, then partitions got the bootloader confused. Another way this can happen is if something during shutdown procedure didn't get transactionally logged. Really hard to debug. Having steps to reproduce is really key.
+    - [Michael, in chat] {%preview https://github.com/ublue-os/bluefin-lts/issues/658 %}
+        - "Apparently the bluefin is related to btrfs + bootc"
+    - [Hristo] Spent a long amount of time trying to reproduce on baremetal machine; in all states, shutdown machine with power switch. Cannot reproduce this issue. Mentioned this in the Fedora discussion thread. But needs that reproduction details. Seems very difficult to reproduce.
+    - [Laura] Can we make a big issue to collect everything?
+    - [Colin] Sure?
+    - [Jorge] Everyone is blaming btrfs.
+    - [Dusty] Also discussed on LinuxUnplugged. Seems to only be on 16.3
+    - **ACTION**: Let's make the catchall here: {%preview https://github.com/ostreedev/ostree/issues/2283 %}
+- ~~[Robert] Systemd Boot - can we start with a hidden flag passed from bootc? https://github.com/coreos/bootupd/pull/978~~ Can move to discussion in the PR
+    - [Laura] Thank you! Please bring it back up again next week if the discussion needs to be live!
+
+### TODO:
+- [ ] Jonathan to make an issue for the plugin discussion to continue offline.
+- [ ] Laura to find out how to get the bootc.dev CNAME setup
+- [ ] All to drop catchall for the persistent bug in the issue
+
diff --git a/meetings/2025-08-22.md b/meetings/2025-08-22.md
@@ -0,0 +1,4 @@
+## 22 Aug 2025
+
+Canceled as many folks have PTO. Enjoy a day off!
+
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,4 @@
		## 22 Aug 2025

		Canceled as many folks have PTO. Enjoy a day off!